WO2017140811A1

WO2017140811A1 - Method and arrangement for calculation of surplus resources for user, frequency or resource trading

Info

Publication number: WO2017140811A1
Application number: PCT/EP2017/053551
Authority: WO
Inventors: Siddarth NAIK; Janis NÖTZEL; Eduard Jorswieck
Original assignee: Technische Universität Dresden
Priority date: 2016-02-16
Filing date: 2017-02-16
Publication date: 2017-08-24
Also published as: EP3417642A1

Abstract

The invention relates to a Method and an Arrangement for Calculation of Surplus Resources for User, Frequency or Resource Trading, in the sense of exchanging resources, in or between mobile network operators. Control units and their operators are to be enabled to achieve the best performance by determining parameters across OSI layers driving the supply and demand functions of resource utilized in radio resource allocation and calculating an individual over- provisioning of resources scenario (OPRS) based on the parameters across OSI layers driving the supply and demand functions of resource utilized in radio resource allocation. An arrangement comprising a control unit is configured for calculation and offering and bidding to control units from other operators for the purpose of identifying surplus and deficit of resources at the local geography.

Description

Method and Arrangement for Calculation of Surplus Resources for User, Frequency or Resource Trading

Technical Field

The invention relates to a Method and an Arrangement for Calculation of Surplus Resources for User, Frequency or Resource Trading in or between mobile network operators.

"Trading" in this invention means primarily the technical possibility to exchange resources.

Background Art

Wireless communication is the backbone of the global economy and the evolution of humanity and artificial intelligence mechanisms.

Spectrum is scarce. Small cell, heterogeneous radio access networks (HetNets) and low latencies require solutions which go beyond pure "traditionally" technical" solutions.

Node to node user trading and resource trading is required and can be achieved via an "enhanced/new" X2 or similar interface .

There are a multitude of parameters which have an effect on the network performance. Certain high level examples of these parameters is as follows: system constants, system parameters, different thresholds and features. Hence, different operators' network do not have similar performances. At certain instances, it could be a trade secret as to which parameters are used by each of the operators. Therefore, there might be an incentive for the operators to buy and sell weakened versions of this information from and to other operators.

A wireless service provider is also known as a mobile network operator, abbreviated as operator which is used in the following. An operator is a provider of services wireless communications that owns or controls all the elements necessary to sell and deliver services to an end user .

Furthermore, it would be very interesting for the vendors to access the performance versus parameter settings of operators and be in a position to get access to the performance of competing vendors and their individual parameter settings.

The objective of the invention is to find a method and arrangement that enables the control units and their corresponding operators to achieve the best performance for themselves and their customer

Nomenclature and Definitions

Speed drop: Drop in the speed for a particular

application running on the end user

equipment, where application is defined as a singular piece of software package (e.g, Netflix application) running on a singular device Cell availability: A key performance indicator (KPI), which displays the percentage of time that a particular cell is available. As for defining the cell as available, it shall be

considered available when the eNodeB can provide E- Radio Access Bearer (E-RAB) service in the cell. time cell is available availability = ^■ 100

measurement time Mobility success rate: A KPI which displays how E-UTRAN mobility functioning is working. The measurement can include both Intra E-UTRAN and Inter RAT

handovers . MobilitySuccessRateq_CI=x = A - B - 100

HO. ExeSucc

A =

HO. ExeAtt

HO. PrepSucc. QCI_QC[=X

B =

HO. PrepAtt. QCI_QCI=x

Where QCI = Quality of Service (QoS) Class Indicator.

The measurement shall include both the preparation and execution phase of the handover. 'Entering preparation phase* is defined as the point of time when the source eNodeB attempts to prepare resources for an user equipment (UE) in a neighboring cell.

'Success of execution phase' is defined as the point of time when the source eNodeB receives information that the UE successfully is connected to the target cell.

RAB retainability is a measurement that shows how often an end-user abnormally loses an E-RAB during the time the E-RAB is used. It is defined as the number of E-RABs with data in a buffer that was abnormally released, normalized with number of data session time units.

IP throughput is a KPI that shows how EUTRAN impact the service quality provided to an end-user.

Payload data volume on IP level pre elapsed time unit on the Uu interface (UMTS air interface, which links User Equipment to the UMTS Terrestrial Radio Access Network, abbr. Uu) . To make sure that only impacts from the RAN is included in this measurement, time units to be included in the elapsed time on the Uu interface shall only be the ones where there is data in the buffer to be

transmitted, e.g. in application data flows such as a web session, there are times when there is no data to transmit by the eNodeB due to bursty traffic pattern, then this eNodeB idle time shall not be included in 'elapsed time unit of the Uu interface'. To achieve a throughput measurement that is independent of the file size it is important to remove the samples where one transmission time interval (TTI) on the radio interface is not utilized.

IP-latency is a measurement that shows how E-UTRAN impacts on the delay experienced by an end-user. Time from reception of IP packet to transmission of the first packet over the Uu . To make sure only contribution from the RAN is included in this measurement, only delay of the first block to the Uu is counted. To achieve a delay

measurement that is independent of IP data block size only the first packet sent to Uu is

measured, To find the delay for a certain packet size the IP throughput measure can be used together with the IP latency (after the first block on the Uu, the remaining time of the packet can be calculated with the IP throughput measure)

CSV Circuit voice CSD circuit switched data

CSV access failure is the product of the RRC connection

failure, NAS setup failure and theRAB establishment failure. Radio resource control (RRC) connection success is counted when the radio network

controller (RNC) received a RRC setup complete for the UE, NAS Setup is considered successful when the signaling messages in the call flow during the call setup flow is successfully

completed by the relevant network elements. A RAB is considered

successfully established when the RAB assignment response is sent by the RNC to the core network (CN) . measures the network's inability to maintain a call. CSV drop is defined as the ratio of abnormal speech disconnects, relative to all speech

disconnects (both normal and abnormal) , A normal disconnect is initiated by a RAB disconnect radio access network application part (RANAP) message from the mobile switching center (MSC) at the completion of the call, An abnormal RAB disconnect includes Radio Link Failures, uplink (UL) or downlink (DL) interference or other reason and can be initiated by either UMTS Terrestrial Radio Access Network (UTRAN) or CN. can be measured by block error rate (BLER) .

CSV quality can be defined separately for downlink and uplink. fter Handover overhead KPI provides an

indication of how many cells or sectors were in the active set during the call on an average basis. CSV Inter-Radio Access Technology ( IRAT) Handover Failure measures the hard handover failure rate across RATs, e.g. UMTS TO GSM' system for voice calls .

CSC call setup time indicates network response time to a user request for a voice service.

CSD access failure is the product of the RRC connection failure, NAS setup failure and the RAB establishment failure. RRC connection success is countered when the RNC receives a RRC setup complete from the UE . NAS Setup is considered successful when the signaling messages in the call flow during call setup flow is

successfully completed by relevant network elements. A RAB is considered to be successfully established when the RAB assignment response is sent by the RNC to the CN.

Packet switched (PS) access failure rate can be measured by the data session (PS) followed by download. In the case of multiple RRC connection requests the first RRC connection will be considered for the KPI calculation. PS data (PSD) drop is considered as dropped when

associated RAB has been released abnormally by either UTRAN or CN. Any drop after RANAP; RAB assignment response is considered as PS drop call.

Similarly, PSD latency, PSD throughput, PSD IRAT Handover Failure, PSD IRAR Interruption Time, High speed downline packet access (HSPDA) access failure, HSDPA drop, HSDPA latency, high speed uplink packet access (HSUPA) throughput can be defined.

Disclosure of Invention

The invention helps a control unit (base station, radio network controller or radio access network) to calculate the deficit resources (e.g. frequencies) it might have based on historical information.

The deficit of resources or the surplus of resources that it might have over a given time horizon for serving users with different service classes could be based on

probability density functions of historical data demand, historical channel conditions, historical user density (in a particular geographic vicinity) , historical mobility and other historical indicators and the correlations in between them. The arrangement helps calculate appropriate over provisioning budgets on each of the parameters at the different layers (open system interface (OSI) model) and on the correlation between them to help the concerned control units to provision the required resources and trade the surpluses with the corresponding control units from the other operators .

Such an over provisioning of engineering units and

parameters has a well-established history and is

frequently practiced in mission critical devices, e.g. transportation bridges, rocket launches.

Furthermore, the arrangement assists in the trading of the probability density functions or the historical time series of the parameters across the OSI layers to assist the nodes of operators to make enhanced radio resource allocation and spectrum management decisions.

It might always be possible for the operator or the node of an operator to mislead another operator or the node of another operator. This misleading could be achieved systematically via programming it into the software or sporadically via other means.

The objective of the invention is solved by a method for calculation of surplus resources comprising

- determining parameters across OSI layers driving the supply and demand functions of resource utilized in radio resource allocation

— calculating individual over-provisioning of resources scenario (OPRS) based on the parameters across OSI layers driving the supply and demand functions of resource utilized in radio resource allocation.

In an embodiment of the invention the parameter are based on the experienced history of the parameters of the OSI model in a given time horizon at a particular geographic location .

The method and arrangement at the control units is able to free up resources for frequency, user and/or resource trading, which could be done via calculating individual over-provisioning of resources scenario (OPRS) based on multiple parameters across the OSI layers driving the supply and demand functions of resource utilized in radio resource allocation, which in turn could be based on the experienced history of the parameters of the OSI model in a given time horizon at a particular geographic location.

It is possible to trade historical time series of

individual factors affect the supply and demand density functions to control units from other operators. The trades could be conducted via a similar interface as that utilized for user and resource trading.

In a further embodiment supply and demand functions are aggregated at a higher hierarchy node (e.g. radio network controller) for resource allocation to the lower hierarchy nodes (e.g. base stations) and resource trading with other radio network controllers.

In one embodiment the OPRS for different supply and demand functions based on the parameters across the OSI layers can be calculated as follows. Let the historical time series for the mobility of a particular user in a

particular geographic vicinity, the historical time series of the demand for a particular service and the channel experienced by this user over a given historical time horizon be available to the arrangement. The arrangement then calculates the right most or the left most tail for a given confidence level (and p-value) such that the

particular parameter is stressed. Namely, left most tail for the channel (worst channels) , right most tails for demand for service and mobility. The method then utilizes the historical worst correlation again with a given confidence interval (and p-value) across these three factors to produce a worst case behavior for this

particular user, when it enters the network. A higher confidence interval leads to taking a smaller mass (area) under the tail.

In one particular embodiment such a generalization of worst case factors could be generalized to a class of users .

It is possible to aggregate supply and demand functions to be used by an operator to optimize resource allocation and investment for each element of the network. Certain examples of these functions are hardware capacity, software capability, base station cell range, number of base stations under an RNC etc.

In one embodiment the method could allow the node to apply a shock (i.e. greater than 100% of the current and

expected value over a time horizon) to a particular supply or demand parameter based on the density function of historical supply or demand (as classified from a resource allocation perspective) . This shock is utilized for over- provisioning for resources either from the resources from one's own network or based on the criticality of the deficit of resources, from another operator's network. In a further embodiment, where a particular operator's node does not have a long enough historical time series for OSI layer parameter (insufficient statistics for this particular parameter) for calculating the OPRS for

calculating a potential and/or expected resource surplus or deficit, the node of this operator could bid for offers of historical time series of this particular supply or demand parameter from the node of another operator in a geographic vicinity. In a further embodiment, where a node of a particular operator has a rich enough time series of all the desired demand and supply parameters, however has not been faced with the problem of aggregating them with a certain correlation between the various parameters at hand, could bid for offers of a correlation figure or a density function of correlations (discretized) from the node of another operator in the geographic vicinity for

calculating of a supply or demand shock for calculating an appropriate deficit or surplus of resources. In case a node of a particular operator has a rich enough time series of all the desired demand and supply

parameters and a set of correlation factors for a set of future time horizons to aggregate these parameters, however there is correlation factor missing for a given time horizon in between the available time horizons the node might not want to utilize interpolation schemes such as cubic splines or higher order polynomials for

calculating the missing correlation factor for the desired time horizon in the future. In this particular case, it has the opportunity to bid for offers of the closest possible correlation coefficient from the nodes of other operators for the desired future time horizon, so as to better aggregate or in the worst case better interpolate before aggregation of the supply and demand shades for calculating a resource surplus or deficit.

In one particular embodiment the node of a particular operator could also be missing an "important" part of the density function (discretized) of the supply or demand parameters, which it can bid for from the nodes of other operators in its geographic vicinity.

It is possible that parameters outside of the OSI layers model, e.g. the prices of resources (e.g. energy

consumption of spectrum consumption) and/or typical key performance indicators (e.g. outage, latency, energy efficiency) could determine the OPRS .

In all the above embodiments, a node of an operator having a very granularity information could also post offers to the nodes of other operators in the geographic vicinity requesting bids from nodes of other operators.

The objective of the invention is also solved by an arrangement for calculation of surplus resources

comprising a control unit in one operator having

connection to control units of other operators. This control unit is configured for calculation and offering and bidding to control units from other operators for the purpose of identifying surplus and deficit of resources at the local geography.

In an embodiment the control unit is configured to calculate appropriate over provisioning budgets on each of the parameters at different layers of an open system interface (OSI) model and on the correlation between them to help the concerned control units to provision the required resources and trade the surpluses with the corresponding control units from the other operators.

Modes for Carrying Out the Invention

In the following the invention is described by examples.

Fig. 1 shows an abstraction of certain supply and demand indicators and their corresponding probability density functions and the correlations between them;

Fig. 2 shows different factors and their individual

periodicities ; Fig. 3 shows an abstraction of protocol of method an

arrangement to calculate surplus resource for user trading and resource trading;

Fig. 4 shows a forecast of the channel realization based on various adverse and less adverse scenarios for calculating a potential deficit and surplus of resources for user and resource trading; and

Fig. 5 shows an abstraction of the choices that the

mechanism (placed at the control unit) has for calculation and offering and bidding to control units from other operators for the purpose of identifying surplus and deficit of resources at the local geography. The method running at the control units could have access to various probability density functions of various factors, This can be achieved by asking the respective radio network controllers for the each of the probability density functions of the supply or demand functions affecting the resources or by accessing the local storage at the control unit itself for a particular geographic boundary .

The supply and demand factors can be divided into the following categories: i. channel quality (path loss, multipath fading and shadowing) ,

ii. achieved combinations of resources (power vectors, beamforming vectors and code books etc.)

iii. mobility functions and

iv. demand functions.

Some of the measurement indicators (across all the OSI layers) that could be picked up to provide an indication of the supply and demand probability demand functions could be classified as follows:

1. Retainabllity of the service.

ii . Utilization of the resources.

ii . Integrity of the service provided,

iv . Mobility of the demand affection the above

factors .

A pictorial description of these probability density functions and the correlations between them has been presented in Fig. 1. In Fig. 1 all probability functions are discrete and are only relevant for a fixed time horizon .

The mechanism could build a forecast of each particular supply and demand factor at a given time horizon t_o based on a history of that particular factor (measurement) from t_-N to t_o.

The control unit could offer this historical time series for this particular measurement to a control unit of another operator via resource trading.

The historical time horizon over which the probability density function of a particular measurement parameter is built could vary from a few micro seconds to a month.

This has been displayed diagrammatically in Fig. 2: i. Second Basis: Fast fading, monetization agreements between content providers and service providers

(for content consumed) .

ii. Minute Basis; Slow fading, schedule of buses,

movement of vehicles post on and off street signals in cities,

iii. Hour Basis; Train moving between cities, plane

movements between cities,

iv. Intra-day Basis: Work cycles of people in a city between residential, office and recreational locations .

v. Daily Basis: Work cycles of people in a city

between residential, office and recreational locations, Sleep rhythm of people in city and in a continent . vi . Weeldy Basis; Week day versus weekend movements, vii. Monthly Basis: Planned festivals and outdoor

events in a city, post salary seasonal activities of people.

viii. Quarterly Basis: New issuance of end-user

equipment, seasonal weather changes

ix. Yearly Basis; Technological developments, new

issuance of infrastructure spots.

In Fig. 2 the different factors and their individual periodicities can be combined such that expectation of supply and demand functions for the future can be built.

The protocol for calculating the surplus resource for user and resource (e.g. frequencies) as part of the method and arrangement could be as displayed in Fig. 3.

The mechanism could forecast a measurement parameter in the future over a given time horizon to to tn for

predefined scenarios. For example in the case of channel quality indicators these forecasting scenarios could be as follows ,

1. Forecast based on severe shadowing.

ii . Forecast based on severe multipath

ii . Forecast based on favorable multipath

iv . Forecast based on no shadowing.

These forecasting scenarios have to be defined explicitly and quantified. A diagrammatic description of the

forecasting scenarios has been presented in Figure 3.

These explicitly defined scenarios could be utilized to define the surplus and demand of the resources to be offered or bid for from the control unit of another operator in the geographical vicinity of the control unit of the operator having made these forecasts.

The mechanism as shown in Fig. 5 could trade the forecasts for a particular measurement parameter to the control unit of another operator in its geographical vicinity via resource trading.

A control unit could specify the scenario that needs to be generated to the control unit of another operator for a particular measurement parameter. Based on the

specification of the scenario the control unit of the later operator could calculate the scenario via the mechanism and provide it to the requesting control unit via resource trading. The mechanism could request correlation values between the measurements parameters such that they can be

appropriately aggregated to get a better perspective on resource allocation. Instead of requesting a scalar correlation value between the probability density

functions of two measurement parameters, the control unit could also request a probability density function of correlation values experiences over a given time horizon at a particular geographic vicinity by the control unit of another operator. In this embodiment the mechanism of a control unit who would have access to such information would build the probability density function and offer it via resource trading to control units of other operators, In reference of this embodiment it should be mentioned that correlation is a linear measure and does not

necessarily measure accurately the mass of the events in the tails. Since the operators and vendors would be more interested in protecting themselves against tail events, other nonlinear measures of dependence between the various measurement parameters could also be traded.

The quality of information provided depends on the number of standard deviations of the probability density function that the mechanism should include to provide the requested data. Hence, the requesting control unit could specify a confidence interval (a percentile value) or a standard deviation value and obtain a price announcement for that information from the control unit of another operator, This price information would correspond to the value corresponding to the standard deviation of the measurement parameter corresponding to a particular announced standard deviation. This has been displayed diagrammatically in Figure 4.

Claims

1. Method for calculation of surplus resources comprising

2. Method according to claim 1, wherein the parameter are based on the experienced history of the parameters of the OSI model in a given time horizon at a particular geographic location.

3. Method according to claim 1 or 2 one embodiment, supply and demand functions are aggregated at a higher

hierarchy node for resource allocation to the lower hierarchy nodes and resource trading with other radio network controllers.

4. Method according to at least one of the claims 1 to 3 wherein the OPRS for different supply and demand functions based on the parameters across the OSI layers are calculated as follows - determining a historical time series for the mobility of a particular user in a particular geographic vicinity, a historical time series of the demand for a particular service and the channel experienced by this user over a given historical time horizon;

- calculating a right most or a left most tail for a given confidence level and p-value such that the particular parameter is stressed, i.e. left most tail for the channel (worst channels) , right most tails for demand for service and mobility;

- producing a worst case behavior for this particular user, when it enters the network by utilizing the historical worst correlation again with a given confidence interval and p-value across these three factors wherein a higher confidence interval leads to taking a smaller mass area under the tail.

Method according to claim 4 wherein a generalization of worst case factors is generalized to a class of users.

Method according to at least one of the claims 1 to 5, wherein a node is allowed to apply a shock to a

particular supply or demand parameter based on the density function of historical supply or demand wherein the shock is utilized for over-provisioning for

resources either from the resources from one's own network or based on the criticality of the deficit of resources from another operator's network.

Method according to at least one of the claims 1 to 6, wherein the node of this operator could bid for offers of historical time series of this particular supply or demand parameter from the node of another operator in a geographic vicinity.

8. Method according to at least one of the claims 1 to 6, wherein a node of a particular operator bids for offers of a correlation figure or a density function of correlations from the node of another operator in the geographic vicinity for calculating of a supply or demand shock for calculating an appropriate deficit or surplus of resources.

9. Method according to at least one of the claims 1 to 8, wherein a node of a particular operator is missing an important part of the density function of the supply or demand parameters, bids for from the nodes of other operators in its geographic vicinity.

10. Method according to at least one of the claims 1 to 8, wherein a node of an operator having a very granularity information post offers to the nodes of other operators in the geographic vicinity requesting bids from nodes of other operators.

11. Arrangement for calculation of surplus resources

comprising a control unit in one operator having connection to control units of other operators

configured for calculation and offering and bidding to control units from other operators for the purpose of identifying surplus and deficit of resources at the local geography.

12. Arrangement according to claim 11 wherein the control unit is configured to calculate appropriate over provisioning budgets on each of the parameters at different layers of an open system interface (OSI) model and on the correlation between them to help the concerned control units to provision the required resources and trade the surpluses with the

corresponding control units from the other operators.